V1.4 Dec 31 2007
This is a work in progress, and it has been a very long time coming. I intend on working on it for the next little while, dumping as much of my brain and hard-earned knowledge into the public record. It is, however, NOT public domain; I reserve the copyright for myself. You may NOT duplicate this elsewhere without my express permission - and this is the age of Google, folks, I WILL find you if you post it elsewhere. Otherwise, read and enjoy, and put this all to good use. Go fast!. DG
This starts with keeping settings as static as possible, which means eliminating flex, binding, squishing, bending, stiffness, hysteresis, asymmetric spring compression, stiction, etc etc etc.
The more force you generate through the tires, the bigger a deal this is. On a street car, you replace any single-axis bushing with poly or nylon, keep the springs coaxial, and call it a day.
Why so much effort on keeping the spring co-axial to the shock? Because any bending moment on the shock rod and spring that occurs when the spring seat doesn't pivot coaxial to the shock rod wears out seals in a hurry, can bend the shock rod, and worst of all, does all kinds of funky things to the spring rates, as the spring is no longer compressed evenly throughout the shock stroke. It is really, really worth the effort to ensure the spring is always compressed squarely.
Most aftermarket suspension providers have figured this out, and even the crappy Japansese and Chinese stuff (JIC, TIEN, etc) comes with coaxial spring perches these days. Some legacy parts (Ground Control in particular) continue to used fixed spring perches, and these should be removed from the car and thrown away with great force.
As the grip goes higher, you need something stiffer than poly, so you do high-quality, Teflon-lined spherical bearings in every pivot point. At still higher grip levels, you start looking at control arm flex and especially chassis flex, and get rid of it as best you can.
Spherical bearings are your friends; the amount of improvement that comes with eliminating rubber can be really suprising. Note, however, that there are different grades of spherical bearing, and there is a signifigant quality difference, for not that big a difference in individual cost. A wide, Teflon-lined, precision-grade bearing is what you want - I used Auroras for the most part. Get the Aurora catalogue; it does a really good job of teaching the engineering required to design parts using spherical bearings.
Note that there is a very large difference between the strength of the bearing loaded axially (like at the top of a shock) and radially (like in a suspension pivot). An axial load takes a much larger bearing than a radial load.
As far as installing bearings goes, pick a bearing from the catalogue, machine out the part to an interference fit (per the specs in the catalogue) - I usually machined out to a shoulder - and retain it with a snap ring for extra insurence. Then you get to make the adaptor sleeves that neck down the bearing ID to the OD of the pivot bolt, and widen the unit up to the width of the attachment clevis.
It can be handy to own a small lathe....
While you're at it, it can be worthwhile to install Torrington bearings in the spring perches:
This gets rid of some stiction at the spring/seat interface (the spring twists as it compresses) and as an added bonus, makes adjusting ride height like 1000% easier. Happily, all you need are two races (the flat plates) and one bearing per spring - no machining required.
At one point, I was putting them on both ends of the spring, but that turns out to be overkill - just one end works fine.
For the ultimate solution, Hypercoil makes floating hydraulic spring seats, but they are pretty expensive and a Torrington bearing with a proper coaxial spring perch gets you most of the way there.
At some point, chassis flex starts becoming a signifigant portion of total suspension movement and until you get rid of that flex further suspension tuning is going to be relatively ineffective. This was a key part of the concept behind Street Modified's "don't touch the chassis" rules and it is suprisingly effective. Look at any high-dollar race series and see how much effort goes into stiffening the chassis....
Now that we have some control over the suspension wandering around on us and muddying up the data, we start looking at how much grip we are making and the resultant roll/pitch angles at max grip. We have to measure/model the suspension to see just how much the design of the suspension is changing the tire's envelope through its range of motion, and if we find changes happening, we have to figure out how to limit them.
In 99% of all cases, this means changing springs. A spring change to a stiffer spring limits the amount of motion that the suspension undergoes in reaction to a particular acceleration (so does a change in CG height and track/wheelbase, but those are much harder to change than springs and for most practical purposes less effective too)
But there are upper limits on how stiff we can go with the springs, so we need a measure of "stiffness" to set the boundaries.
That number is the natural frequency of the suspension. I'm not going to describe how to calculate it just yet; look it up for now. Rule of thumb is rear NF slightly higher than front (by a tenth of a Hz or two) Street car: 0.8 Hz. Occasional autocrosser: 1-1.5 Hz. Full-bore autocrosser: 2.2-2.5 Hz.
Yes, it really is that simple. Measure your corner weights, unsprung masses, and motion ratios, and then pick springs that put the front NF at 2.2 Hz and the rear at 2.5 Hz.
Of all the claims I've made in this book, this is probably the one that caused the most serious discussion. It's also the one that is the hardest to justify, given that it is a bit of a "magic number" and tire and surface dependant at that. For a slicker or bumpier surface, it could probably drop a tenth of a Hz or so. Grippier tires (like slicks - although sidewall stiffness will play a part here) might jack it up a little.
The point here isn't if the car should be 2.2000 Hz or 2.201324 Hz - the major takeaway is that ~2.2 Hz is the ballpark. If the natural frequencies on your Camaro work out to 1.7 Hz front and 3.9 Hz rear... well, you've got a problem there Skippy. (A car like this should be undrivable, all loosey-goosey-can't-put-power-down - unless the front shocks have massive amounts of rebound to compensate....)
Note too that, very much unlike spring rates, natural frequencies CAN be compared car-to-car. A 96 Camaro with a coaxial McStrut (motion ratio ~ .98) running a 700lb spring has a similar natural frequency to a 96 Mustang with an A-Arm mounted spring (motion ratio ~ .4) and a 1600lb spring. Way different spring rates, nearly identical natural frequencies.
While we're on the topic of "picking springs", I am of the opinion that there is only one spring provider worth a damn - Hypercoil. Of all the springs I have tested - you need a spring tester, MK Technologies makes a good one (and Hoosier Tom will sell it to you) - the only springs that reliably tested in spec were the Hypercoils. The worst offenders were the no-name purple springs that come on JICs; I saw 25% variation on some of these.
Just like shocks you must test your springs, and do so regularly. The spec on a Hypercoil is 3%, and at stiffer rates, it is possible to try and add 50lbs of rate, but swap in a spring on the soft side of the spec for a spring on the stiff side of the spec, and effectively do nothing except fool yourself. TEST TEST TEST! Unlike shocks, no spring vendor is going to provide test results for individual springs, so you get to buy a spring tester. Tell Marty and Tom I said "Hi".
I'm also a huge fan of using standard 2.5" or 2.25" springs in a coilover configuration whenever possible. Not only does this package beautifully, it gives you access to a huge, fine-grained range of spring rates to choose from. A Hypercoil 2.25" spring is THE quality bang for the buck deal in racing; they could be charging well in excess of what they do. Not only that, a coaxial coilover configuration means the springs and shocks have the same motion ratio, which makes the NF calculations a little bit easier.
If you can, use the smaller 2.25" springs - they are a little bit lighter and give a touch more clearence around the shock.
Now you have the operating envelope of the suspension, so you reset all the suspension curves to be as flat as possible over this range of motion, particularly bump steer and camber change. Lower the suspension as much as you can while keeping the curves sane, and if you have to trade a higher CG for better camber curves, go with the better camber curves.
So then, here's the sequence: